Chapter 28:
IBM DB2 Universal Database
Database System Concepts, 5th Ed.
©Silberschatz, Korth and Sudarshan
Database System Concepts






Chapter 1: Introduction
Part 1: Relational databases

Chapter 2: Relational Model

Chapter 3: SQL

Chapter 4: Advanced SQL

Chapter 5: Other Relational Languages
Part 2: Database Design

Chapter 6: Database Design and the E-R Model

Chapter 7: Relational Database Design

Chapter 8: Application Design and Development
Part 3: Object-based databases and XML

Chapter 9: Object-Based Databases

Chapter 10: XML
Part 4: Data storage and querying

Chapter 11: Storage and File Structure

Chapter 12: Indexing and Hashing

Chapter 13: Query Processing

Chapter 14: Query Optimization
Part 5: Transaction management

Chapter 15: Transactions

Chapter 16: Concurrency control

Chapter 17: Recovery System
Database System Concepts - 5th Edition, May 23, 2005





Part 6: Data Mining and Information Retrieval

Chapter 18: Data Analysis and Mining

Chapter 19: Information Retreival
Part 7: Database system architecture

Chapter 20: Database-System Architecture

Chapter 21: Parallel Databases

Chapter 22: Distributed Databases
Part 8: Other topics

Chapter 23: Advanced Application Development

Chapter 24: Advanced Data Types and New Applications

Chapter 25: Advanced Transaction Processing
Part 9: Case studies

Chapter 26: PostgreSQL

Chapter 27: Oracle

Chapter 28: IBM DB2

Chapter 29: Microsoft SQL Server
Online Appendices

Appendix A: Network Model

Appendix B: Hierarchical Model

Appendix C: Advanced Relational Database Model
27.2
©Silberschatz, Korth and Sudarshan
Part 9: Case studies
(Chapters 26 through 29).
 Chapter 26: PostgreSQL
 Chapter 27: Oracle
 Chapter 28: IBM DB2
 Chapter 29: Microsoft SQL Server.
 These chapters outline unique features of each of these systems, and describe
their internal structure.
 They provide a wealth of interesting information about the respective products, and
help you see how the various implementation techniques described in earlier parts
are used in real systems.
 They also cover several interesting practical aspects in the design of real systems.
Database System Concepts - 5th Edition, May 23, 2005
27.3
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.4
©Silberschatz, Korth and Sudarshan
Overview
 The origin of DB2

The System R project at IBM’s Almaden Research Center in 1976
 The first DB2 product was released in 1984
 IBM continually enhanced the DB2 product in areas

Transaction processing

Query processing and optimization

Parallel processing

Active database support

Advanced query and warehousing techniques

Object-relational support
Database System Concepts - 5th Edition, May 23, 2005
27.5
©Silberschatz, Korth and Sudarshan
Overview (cont.)
 DB2 database engine consists of four code base types

Linux, Unix, and Windows

z/OS

VM

OS/400
 In this chapter, the focus is on the DB2 Universal Database (UDB)
engine that supports Linux, Unix, and Windows
 The latest version of DB2 is version 8.2, which improves the scalability,
availability, and general robustness of the DB2 engine
Database System Concepts - 5th Edition, May 23, 2005
27.6
©Silberschatz, Korth and Sudarshan
Overview (cont.)
Database System Concepts - 5th Edition, May 23, 2005
27.7
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.8
©Silberschatz, Korth and Sudarshan
Database-Design Tools
 DB2 provides support for many logical and physical database features
using SQL
 The feature includes constraints, triggers, and recursion using SQL
constructs
 Physical database features such as tablespaces, bufferpools, and
partitioning are also supported by using SQL statements
 DB2 Control Center

Includes various design- and administration-related tools

Provides a tree-view of a server and its database

Allows users to define new objects, create ad-hoc SQL queries,
and view query results
Database System Concepts - 5th Edition, May 23, 2005
27.9
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.10
©Silberschatz, Korth and Sudarshan
SQL Variations and Extensions
DB2 supports a rich set of SQL features for database processing
 XML features
A rich set of XML functions have been included in DB2

xmlelement

xmlattributes

xmlforest

xmlconcat

xmlserialize

xmlagg

xml2clob
 Fig. 28.2 shows an SQL query with XML extensions
 Fig. 28.3 shows its resultant output
Database System Concepts - 5th Edition, May 23, 2005
27.11
©Silberschatz, Korth and Sudarshan
SQL Variations and Extensions (cont.)
Database System Concepts - 5th Edition, May 23, 2005
27.12
©Silberschatz, Korth and Sudarshan
SQL Variations and Extensions (cont.)
 Support for Data Types

DB2 supports user-defined data types

Users can define distinct or structured data types

User-defined data types: distinct
create distinct type us_dollar as decimal(9,2)

User can create a field in a table with type us_dollar
select product from us_sales
where price > us_dollar(1000)
Database System Concepts - 5th Edition, May 23, 2005
27.13
©Silberschatz, Korth and Sudarshan
SQL Variations and Extensions (Cont.)
 Support for Data Types (cont.)

User-defined data types: structured
Structured data types are complex objects that usually consist of
two or more attributes.
create type department_t as
(deptname varchar(32),
depthead varchar(32),
faculty_count integer)
mode db2sql

Structured types can be used to define typed tables
create table dept of department_t
Database System Concepts - 5th Edition, May 23, 2005
27.14
©Silberschatz, Korth and Sudarshan
SQL Variations and Extensions (Cont.)
 User-Defined Functions (UDFs) and Methods

Users can define their own functions and methods

Functions can generate scalars (single attribute) or tables
(multiattribute row) as their result

User can register functions using the create function statement

User-defined functions can operate in fenced or unfenced modes.
 In
fenced mode: the functions are executed by a separate thread
in its own address space
 In
unfenced mode: the database processing agent is allowed to
execute the function in the server’s address space
Database System Concepts - 5th Edition, May 23, 2005
27.15
©Silberschatz, Korth and Sudarshan
SQL Variations and Extensions (Cont.)
 User-Defined Functions (UDFs) and Methods (cont.)

Fig 28.4 shows a definition of UDF, db2gse.GsegeFilterDist
Database System Concepts - 5th Edition, May 23, 2005
27.16
©Silberschatz, Korth and Sudarshan
SQL Variations and Extensions (Cont.)
 Large Objects

New database applications require the manipulation of text,
images, video, and other types of data that are typically quite large

DB2 supports these requirements by providing three different large
object (LOB) types

Binary Large Objects (blobs)

Single Byte Character Large Objects (clobs)

Double Byte Character Large Objects (dbclobs)
Database System Concepts - 5th Edition, May 23, 2005
27.17
©Silberschatz, Korth and Sudarshan
SQL Variations and Extensions (Cont.)
 Indexing Extensions and Constraints

Users can create index extensions to generate keys from
structured data types by using create index extension statement
Database System Concepts - 5th Edition, May 23, 2005
27.18
©Silberschatz, Korth and Sudarshan
SQL Variations and Extensions (Cont.)
 Web Services

DB2 can integrate Web services as producer or consumer

For example, a Web service invokes the following SQL
select trn_id, amount, date
from transactions
where cust_id = <input>
order by date
fetch first 1 row only;

The following SQL shows DB2 acting as a consumer of a Web
service
select ticker_id, GetQuote(ticker_id)
from portfolio
Database System Concepts - 5th Edition, May 23, 2005
27.19
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.20
©Silberschatz, Korth and Sudarshan
Storage and Indexing
 Storage Architecture

DB2 provides storage abstraction for managing logical database
tables usefully in multinode and multidisk environment

Nodegroups can be defined to support table partitioning across a
specific set of nodes in a multinode system
→ Flexibility in allocating table partitions
ex) Large tables may be partitioned across all nodes in a system
Small tables may reside on a single node
Database System Concepts - 5th Edition, May 23, 2005
27.21
©Silberschatz, Korth and Sudarshan
Storage and Indexing (cont.)
 Storage Architecture (cont.)

DB2 uses tablespaces to organize tables

A tablespace consists of one or more containers, which are
references to directories, devices, or files

A tablespace may contain zero or more database objects such as
tables, indices, or LOBs

Fig 28.6 shows that

Two tablespaces are defined for a nodegroup

Humanres tablespace is assigned four containers

Sched tablespace has only one container

Employee and department tables are assigned to the humanres
tablespace

Project table is in the sched tablespace
Database System Concepts - 5th Edition, May 23, 2005
27.22
©Silberschatz, Korth and Sudarshan
Storage and Indexing (cont.)
Database System Concepts - 5th Edition, May 23, 2005
27.23
©Silberschatz, Korth and Sudarshan
Storage and Indexing (cont.)
 Storage Architecture (cont.)

Two kinds of tablespaces

System-managed spaces (SMS)
: Directories or file systems that the underlying operating system
maintains

Data-managed spaces (DMS)
: Raw devices or preallocated files that are then controlled by DB2

DB2 support stripping across the different containers

One-by-one extent page in round-robin fashion

Benefits are parallel I/O and load balancing
Database System Concepts - 5th Edition, May 23, 2005
27.24
©Silberschatz, Korth and Sudarshan
Storage and Indexing (cont.)
 Buffer Pools

A buffer pool is a common shared data area that maintains
memory copies of objects

One or more buffer pools may be associated each tablespace for
managing different objects

DB2 allows buffer pools to be defined by SQL statements
create bufferpool <buffer-pool> ….
alter bufferpool <buffer-pool> size <n>
Database System Concepts - 5th Edition, May 23, 2005
27.25
©Silberschatz, Korth and Sudarshan
Storage and Indexing (cont.)
 Tables, Records, and Indices

A table consists of a set of pages, and each page consists of a
set of records

user data records or special system records

DB2 uses a space map record called Free Space Control Record
(FSCR) to find free space in table

Fig 28.7 shows the logical view of a table and an associated
index
Database System Concepts - 5th Edition, May 23, 2005
27.26
©Silberschatz, Korth and Sudarshan
Storage and Indexing (cont.)
Database System Concepts - 5th Edition, May 23, 2005
27.27
©Silberschatz, Korth and Sudarshan
Storage and Indexing (cont.)
 Tables, Records, and indices (cont.)

Indices are organized as pages containing index records and
pointers to child and sibling pages

DB2 supports B+-tree index mechanisms


DB2 contains internal pages and leaf pages

Leaf pages contain index entries that point to records in the table

Each record in a table can be uniquely identified by using its page and
slot information, called a record ID (RID)
DB2 supports “include columns” in the index definition
create unique index I1 on T1(C1) include (C2)

Specifies that C2 is to be included as an extra column in an index on
column C1

This enables DB2 use “index-only” query processing techniques
Database System Concepts - 5th Edition, May 23, 2005
27.28
©Silberschatz, Korth and Sudarshan
Storage and Indexing (Cont.)
 Tables, Records, and indexes (cont.)

Data page
Fig. 28.8 shows the typical data page format in DB2

Each data page contains a header and a slot directory

The slot directory is an array of 255 entries that points to record offsets
in the page

Page 1056 contains record 1 at offset 3700, which is a forward pointer
to the record <473, 2>

Record <473,2> is an overflow record created as a result of an update
operation of the original record <1056,1>

DB2 supports different page sizes such as 4,8,16 and 32 KB
Database System Concepts - 5th Edition, May 23, 2005
27.29
©Silberschatz, Korth and Sudarshan
Storage and Indexing (cont.)
Database System Concepts - 5th Edition, May 23, 2005
27.30
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.31
©Silberschatz, Korth and Sudarshan
Multidimensional Clustering
 Multidimensional Clustering (MDC)

DB2 table may be created by specifying one or more keys as
dimensions along which to cluster the table’s data

A new clause called organize by dimensions

For example, the following DDL describes a sales table organized
by storedId, year(orderDate), and itemId as dimensions
create table sales(storeId int,
orderDate date,
shipDate date,
receiptDate date,
region int,
itemId int,
price float,
yearOd int generated always as year(orderDate))
organize by dimensions (region, yearOd, itemId)
Database System Concepts - 5th Edition, May 23, 2005
27.32
©Silberschatz, Korth and Sudarshan
Multidimensional Clustering (cont.)
 Multidimensional Clustering (MDC) (cont.)

Each of dimensions may consist of one or more columns

A ‘dimension block index’ is

automatically created for each of the specified dimensions

used to access data quickly and efficiently

A composite block index is used to maintain the clustering of data
over insert and update activity

Every unique combination of dimension values forms a logical
“cell”, which is physically organized as block of pages

Fig 28.9 shows a simple logical cube with only two values for each
dimension attribute
Database System Concepts - 5th Edition, May 23, 2005
27.33
©Silberschatz, Korth and Sudarshan
Multidimensional Clustering (cont.)
Database System Concepts - 5th Edition, May 23, 2005
27.34
©Silberschatz, Korth and Sudarshan
Multidimensional Clustering (cont.)
 Block Indices

A dimension block index is created on each of the year(orderDate),
region, and itemId attributes

Each is structured in the same manner as a traditional B-tree index


except that, at the leaf level, the keys point to a block identifier
(BID) instead of a RID
Fig. 28. 10 illustrates slices of blocks for specific values of region
and itemId dimensions, respectively
Database System Concepts - 5th Edition, May 23, 2005
27.35
©Silberschatz, Korth and Sudarshan
Multidimensional Clustering (cont.)
Database System Concepts - 5th Edition, May 23, 2005
27.36
©Silberschatz, Korth and Sudarshan
Multidimensional Clustering (cont.)
 Block Map

A block map

Is also associated with the table

Records the state of each block belonging to the table

May be in such states as
– In use
– Free
– Loaded
– Requiring constant enforcement
Database System Concepts - 5th Edition, May 23, 2005
27.37
©Silberschatz, Korth and Sudarshan
Multidimensional Clustering (cont.)
 Design Considerations


A crucial aspect of MDC is to choose

The right set of dimensions for clustering a table

The right block size parameter to minimize the space utilization
If chosen appropriately,


If chosen incorrectly,


Significant performance and maintenance advantages
Performance degradation and worse space utilization
There are a number of tuning knobs that can be exploited to
organize the table
Database System Concepts - 5th Edition, May 23, 2005
27.38
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.39
©Silberschatz, Korth and Sudarshan
Query Processing and Optimization

DB2 queries are transformed into a tree of operations by the query
compiler

The query operator tree is used at execution time for processing

DB2 supports a rich set of query operators

Figure 28.12 and 28.13 show a query and its associated query plan

The query is a representative complex query from the TPC-H benchmark
and contains several joins and aggregations

The query plan in this example is rather simple

DB2 provides powerful visual explain feature that can help users
understand the details of a query execution plan

The query plan in the figure is based on the visual explain for the query
Database System Concepts - 5th Edition, May 23, 2005
27.40
©Silberschatz, Korth and Sudarshan
Query Processing and Optimization (cont.)
Database System Concepts - 5th Edition, May 23, 2005
27.41
©Silberschatz, Korth and Sudarshan
Query Processing and Optimization (cont.)
# TPC- H benchmark

The TPC is a non-profit corporation founded to define transaction
processing and database benchmarks and to disseminate objective,
verifiable TPC performance data to the industry

The TPC Benchmark (TPC-H) is a decision support benchmark

It consists of a suite of business oriented ad-hoc queries and
concurrent data modifications

The queries and the data populating the database have been chosen
to have broad industry-wide relevance

This benchmark illustrates decision support systems that examine
large volumes of data, execute queries with a high degree of
complexity, and give answers to critical business questions
Database System Concepts - 5th Edition, May 23, 2005
27.42
©Silberschatz, Korth and Sudarshan
Query Processing and Optimization (cont.)
 Access Methods
DB2 supports a comprehensive set of access methods

Table scan
 Is
the most basic method and performs a page-by-page access
of all records in the table

Index scan
 DB2

accesses the qualifying records using the RIDs in the index
Block index scan
 A new
 The
access method for MDC tables
qualifying blocks are accessed and processed
Database System Concepts - 5th Edition, May 23, 2005
27.43
©Silberschatz, Korth and Sudarshan
Query Processing and Optimization (cont.)
 Access Methods (cont.)

Index only
 Is
used when the index contains all the attributes required
 A scan
of the index entries is sufficient, and there is no need to
fetch the records

List prefetch
 This
access method is chosen for an unclustered index scan with
a significant number of RIDs
 DB2
collects the RIDs of relevant records using an index scan
 Then
sorts the RIDs by page number
 Finally
performs a fetch of the records in sorted order from the
data pages
Database System Concepts - 5th Edition, May 23, 2005
27.44
©Silberschatz, Korth and Sudarshan
Query Processing and Optimization (cont.)
 Access Methods (cont.)


Block and record index ANDing

DB2 uses this method when it determines that more than one index
can be used to constrain the number of satisfying records in a base
table

It processes the most selective index to generate a list of BIDs or RIDs

It then processes the next selective index to return the BIDs or RIDs

A BID or RID qualifies for further processing only if it is present in the
intersection (AND operation)
Block and record index ORing

Two or more block or record indices can be used to satisfy query
predicates that are combined by using the OR operator

DB2 eliminates duplicate BIDs or RIDs by performing a sort and then
fetches the resulting set of records
Database System Concepts - 5th Edition, May 23, 2005
27.45
©Silberschatz, Korth and Sudarshan
Query Processing and Optimization (cont.)
 Join, Aggregation, and Set Operations

DB2 can choose between nested-loop, sort-merge, and hash-join
techniques

In describing the join and set binary operation, we use the notation
of “outer” and “inner” tables to distinguish the two input streams

Nested-loop technique is useful if the inner table
– Is very small
– Or can be accessed by using an index on a join predicate

Sort-merge-join and hash-join techniques are used for joins involving
large outer and inner tables

Merging technique eliminates duplicates in the case of union

DB2 implements set operations by using sorting and merging
techniques
Database System Concepts - 5th Edition, May 23, 2005
27.46
©Silberschatz, Korth and Sudarshan
Query Processing and Optimization (cont.)
 Support for Complex SQL Processing

One of the most important aspects of DB2 is that it uses the query
processing infrastructure in an extensible fashion to support
complex SQL operation

The complex SQL operations include support for deeply nested
and correlated subqueries as well as constraints, referential
integrity, and triggers

DB2 can provide better efficiency and scalability by performing
most of the constraint checking and triggering actions as part of the
query plan
Database System Concepts - 5th Edition, May 23, 2005
27.47
©Silberschatz, Korth and Sudarshan
Query Processing and Optimization (cont.)
 Multiprocessor Query-Processing Features

DB2 extends the base set of query operation with control and data
exchange primitives to support SMP, MPP, and SMP-cluster modes
of query processing

DB2 uses a “tablequeue” abstraction for data exchange between
threads on different nodes or on the same node

The table queue is a buffer that redirects data to appropriate
receivers by using broadcast, one-to-one, or directed multicast
methods

In all these modes, DB2 employs a coordinator process to control
the query operations and final result gathering

Coordinator processes can also perform some global databaseprocessing actions
Database System Concepts - 5th Edition, May 23, 2005
27.48
©Silberschatz, Korth and Sudarshan
Query Processing and Optimization (cont.)
 Multiprocessor Query-Processing Features (cont.)

Fig 28.14 shows a simple query executing in a 4-node MPP system
Database System Concepts - 5th Edition, May 23, 2005
27.49
©Silberschatz, Korth and Sudarshan
Query Processing and Optimization (cont.)
 Query Optimization
Procedures

Parsing the SQL statement to QGM (Query Graph Model)

Performing semantic transformations on the QGM to enforce
constraints, referential integrity, and triggers → Enhanced QGM

Performing query rewrite transformations


Decorrelation of correlated subqueries

Transformation of subqueries into joins using early-out processing

Pushing the group by operation below joins

Using materialized view for portions of the original query
The query optimizer component uses this enhanced and
transformed QGM as its input for optimization
Database System Concepts - 5th Edition, May 23, 2005
27.50
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.51
©Silberschatz, Korth and Sudarshan
Materialized Query Tables
 Materialized Query Tables (MQTs)

In DB2, the materialized views are called materialized query tables

MQTs are specified by using create table statement as shown by
Fig. 28.15
Database System Concepts - 5th Edition, May 23, 2005
27.52
©Silberschatz, Korth and Sudarshan
Materialized Query Tables (cont.)
 Query Routing to MQTs


The internal QGM model allows
the compiler to

match the input query against
the available MQT definitions

choose appropriate MQTs for
consideration
Fig. 28.16 shows the entire flow
of the reroute and optimization
Database System Concepts - 5th Edition, May 23, 2005
27.53
©Silberschatz, Korth and Sudarshan
Materialized Query Tables (cont.)
 Maintenance of MQTs

Two dimensions to maintenance: time and cost
 Time
dimension
– Immediate
– Deferred
 Size
dimension
– Incremental
– Full

Fig. 28.17 shows the two dimensions and the options that are the
most useful
Database System Concepts - 5th Edition, May 23, 2005
27.54
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.55
©Silberschatz, Korth and Sudarshan
Autonomic Features in DB2
 Autonomic computing includes a set of techniques that allow the
computing environment to manage itself and reduce the external
dependencies
 Configuration

DB2 supports autonomic tuning of various memory and system
configuration parameters

For instance, parameters such as buffer pool sizes and sort heap
sizes can be specified as autonomic
Database System Concepts - 5th Edition, May 23, 2005
27.56
©Silberschatz, Korth and Sudarshan
Autonomic Features in DB2 (cont.)
 Optimization

Important aspects of improving the performance
 Auxiliary
 Data

data structures (indices, MQTs)
organization features (partitioning, clustering)
Design Advisor
 Provides
workload-based advice
 Automatically

analyze a workload, using optimization techniques
ex) db2advis -d <DB name> -i <workloadfile> -m MICP
– M: Materialized query tables
– I: Indices
– C: Clustering, namely, MDC
– P: Partitioning key selection
Database System Concepts - 5th Edition, May 23, 2005
27.57
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.58
©Silberschatz, Korth and Sudarshan
Tools and Utilities
 The DB2 control center is the primary tool for use and administration of
DB2 databases
 It is organized from data objects such as servers, databases, tables,
and indices
 It contains task-oriented interfaces to perform commands allows users
to generate SQL scripts
 Fig. 28.18 shows a screen shot of the main panel of the Control Center
Database System Concepts - 5th Edition, May 23, 2005
27.59
©Silberschatz, Korth and Sudarshan
Tools and Utilities (cont.)
Database System Concepts - 5th Edition, May 23, 2005
27.60
©Silberschatz, Korth and Sudarshan
Tools and Utilities (cont.)
 Utilities

DB2 provides comprehensive support for load, import, export, reorg,
redistribute, and other data-related utilities

DB2 supports a number of tools such as
 Audit
facility
 Governor
facility
 Query
patroller facility
 Trace
and diagnostic facilities
 Event
monitoring facilities
Database System Concepts - 5th Edition, May 23, 2005
27.61
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.62
©Silberschatz, Korth and Sudarshan
Concurrency Control and Recovery
 Concurrency and Isolation

DB2 supports various modes of concurrency control and isolation

Repeatable read (RR) mode
 Corresponds
closely to the serializable level
 Ensures
that range scans will find the same set of tuples if
repeated

Read stability (RS) mode


Cursor stability (CS) mode


Locks only the rows that an application retrieves in a unit of work
Default isolation level
Uncommitted read (UR) mode
Database System Concepts - 5th Edition, May 23, 2005
27.63
©Silberschatz, Korth and Sudarshan
Concurrency Control and Recovery (cont.)
 Concurrency and Isolation (cont.)

The various isolation modes are implemented by using locks

DB2 supports record-level and table-level locks

DB2 escalates from record-level to table-level locks if the space in
the lock table becomes tight

DB2 supports a variety of lock modes in order to maximize
concurrency

Fig. 28.19 shows the different lock modes and their descriptions
Database System Concepts - 5th Edition, May 23, 2005
27.64
©Silberschatz, Korth and Sudarshan
Concurrency Control and Recovery (cont.)
Database System Concepts - 5th Edition, May 23, 2005
27.65
©Silberschatz, Korth and Sudarshan
Concurrency Control and Recovery (cont.)
 Logging and Recovery


DB2 supports two types of log modes

Circular logging : useful for crash recovery or application failure
recovery

Archival logging : required for roll-forward recovery from an archival
backup
DB2 supports transaction rollback and crash recovery as well as
point-in-time or roll-forward recovery

In crash recovery, DB2 performs the standard phases of undo and redo
processing

For point-in-time recovery, database can be restored from a backup and
can be rolled forward to a specific point in time using the archived logs

The roll-forward recovery command supports both database and
tablespace levels
Database System Concepts - 5th Edition, May 23, 2005
27.66
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.67
©Silberschatz, Korth and Sudarshan
System Architecture
 DB2 Process Model
Fig. 28.20 shows some of the different process or threads in DB2

Remote client applications connect to the database server through
communication agents such as db2tcpcm

Each application is assigned an agent called the db2agent thread

A dedicated server process that serves for each active connection

This agent and its subordinate agents perform the applicationrelated tasks

There are a set of agents at the level of the server to perform tasks
such as crash detection, license server, process creation and
control of system resources
Database System Concepts - 5th Edition, May 23, 2005
27.68
©Silberschatz, Korth and Sudarshan
System Architecture (Cont.)
Database System Concepts - 5th Edition, May 23, 2005
27.69
©Silberschatz, Korth and Sudarshan
System Architecture (Cont.)
 DB2 Memory Model
Fig. 28.21 shows the different types of memory segment in DB2

Private memory in agents or threads is mainly used for local
variables and data structures that are relevant only for the current
activity

Shared memory is partitioned into
 database
shared memory
– contains useful data structures such as buffer pool, lock lists,
application package caches, and shared sort areas
 server
shared memory and application shared memory
– are primarily used for common data structures and
communication buffers
Database System Concepts - 5th Edition, May 23, 2005
27.70
©Silberschatz, Korth and Sudarshan
System Architecture (Cont.)
Database System Concepts - 5th Edition, May 23, 2005
27.71
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.72
©Silberschatz, Korth and Sudarshan
Replication, Distribution, and External Data
 DB2 Replication
A product in the DB2 family that provides replication capabilities
among DB2 other relational data sources such as Oracle, MS
SQL Server, and other nonrelational data sources
Queue Replication
 Creates a queue transport mechanism using IBM’s messagequeue product
DB2 Information-integrator product
 Provides federation, replication, and search capabilities
User-defined table functions
 Enable access to nonrelational and external data sources
 create function statement with the clause returns table
Two-phase commit protocol
 Provides full supports for distributed transaction processing





Database System Concepts - 5th Edition, May 23, 2005
27.73
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.74
©Silberschatz, Korth and Sudarshan
Business Intelligence Features
 DB2 Data Warehouse Edition

An offering that incorporates business intelligence features

Enhances the DB2 engine with features for ETL, OLAP, mining,
and on-line reporting
 On-line Analytical Processing (OLAP)

Cube views provides a mechanism to construct appropriate data
structures and MQTs

DB2 also provides multidimensional OLAP support using DB2
OLAP server

DB2 Alphablox is a new feature that provides on-line, interactive,
reporting, and analysis capabilities

DB2 Intelligent Miner provides various components for modeling,
scoring, and visualizing data
Database System Concepts - 5th Edition, May 23, 2005
27.75
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Summary & Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.76
©Silberschatz, Korth and Sudarshan
Summary (1)
 This chapter provides a very brief synopsis of the features that are
available in the DB2 Universal Database System
 In short, the DB2 Universal Database is multi-platform, scalable,
object-relational database server
 DB2 Control Center includes various design- and administration-
related tools
 DB2 provides support for a rich set of SQL features for various aspects
of database processing
 The storage and indexing architecture in DB2 consists of the file-
system or disk-management layer, the services to manage the buffer
pools, data objects, and concurrency and recovery managers
Database System Concepts - 5th Edition, May 23, 2005
27.77
©Silberschatz, Korth and Sudarshan
Summary (2)
 DB2 provides multidimensional clustering (MDC)
 Materialized views are supported in DB2
 DB2 UDB version 8.2 provides features for simplifying the design and
manageability of databases
 DB2 provides a number of tools for ease of use and administration
 DB2 supports a comprehensive set of concurrency-control, isolation,
and recovery techniques
 DB2 Data Warehouse Edition is an offering in the DB2 family that
incorporates business intelligence features
Database System Concepts - 5th Edition, May 23, 2005
27.78
©Silberschatz, Korth and Sudarshan
Bibliographical Notes (1)
 The origin of DB2 can be traced back to the System R Project (Chamberlin et
al.[1981])
 IBM Research contributions include areas such as transaction processing
(write-ahead logging and ARIES recovery algorithms) (Mohan et al.[1992]),
query processing optimization (Starbust) (Haas et al.[1990]), parallel
processing (DB2 Parallel Edition) (Baru et al.[1995]), active database support
(constraints, triggers) (Cochrane et al.[1996]), advanced query and
warehousing techniques such as materialized views (Zaharioudakis et
al.[2000], Lehner et al.[2000]), multidimensional clustering (Padmanabahn et
al.[2003], Bhattacharjee et al.[2003]), autonomic features (Zilio et al.[2004])
and object-relational support (ADTs, UDFs) (Carey et al.[1999])
 Multiprocessor-query-processing details can be found in Baru et al.[1995] or in
the DB2 Administration and Performance guides DB2 Online documentation
Database System Concepts - 5th Edition, May 23, 2005
27.79
©Silberschatz, Korth and Sudarshan
Bibliographical Notes (2)
 Don Chamberlin’s books provide a good review of the SQL and programming
features (Chamberlin[1996] and Chamberlin[1998])
 Earlier books by C. J. Date and others provide a good review of the features of
DB2 Universal Database for OS/390 (Date[1989] and Martin et al.[1989])
 Recent book such as DB2 for Dummies (Zikopoulos et al.[2000]), DB2 SQL
developer’s guide (Sanders[2000]), and the DB2 Administration Certification
Guides (Cook et al.[1999]) provide hand-on training for using and
administering DB2
 Chamberlin[1998], Zikopoulos et al.[2004] and the DB2 documentation library
provide a complete description of the SQL support
Database System Concepts - 5th Edition, May 23, 2005
27.80
©Silberschatz, Korth and Sudarshan
Chapter 28: IBM DB2 Universal Database
 28.1 Overview
 28.2 Database-Design Tools
 28.3 SQL Variations and Extensions
 28.4 Storage and Indexing
 28.5 Multidimensional Clustering
 28.6 Query Processing and Optimization
 28.7 Materialized Query Tables
 28.8 Autonomic Features in DB2
 28.9 Tools and Utilities
 28.10 Concurrency Control and Recovery
 28.11 System Architecture
 28.12 Replication, Distribution, and External Data
 28.13 Business Intelligence Features
 Summary & Bibliographical Notes
Database System Concepts - 5th Edition, May 23, 2005
27.81
©Silberschatz, Korth and Sudarshan
End of Chapter
Database System Concepts, 5th Ed.
©Silberschatz, Korth and Sudarshan